JP2002353107A - Analyzing method and apparatus in wafer manufacturing process - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an analyzing method and an analyzing apparatus, in a wafer manufacturing process which contribute to improvement of manufacturing processes, by systematically and effectively analyzing the fault pattern of wafers. SOLUTION: According to an analyzing method and an analyzing apparatus in the wafer-manufacturing process, a plurality of intrinsic distribution patterns are defined; a plurality of distribution characteristics having correlation with at least one pattern of a plurality of intrinsic distribution patterns is defined; each pattern group of wafer is caused to have correlation to at least one pattern of a plurality of intrinsic distribution patterns; and each fault pattern distribution pattern is caused to have indirect correlation to at least one distribution characteristic, based on the correlation between the distribution characteristic and intrinsic distribution patterns and the correlation between the fault pattern distribution pattern and intrinsic distribution pattern. The amount of influence of the relevant distribution characteristic can be obtained by respectively setting the correlation coefficients and accumulating all correlation coefficients, corresponding to the relevant distribution characteristic of each distribution characteristic.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an analysis method and apparatus in a wafer manufacturing process, and more particularly, to analyzing and classifying a defect pattern group of a wafer in a semiconductor manufacturing process to obtain information contributing to improvement of the manufacturing process. And a method and apparatus therefor.

[0002]

2. Description of the Related Art Semiconductor parts are manufactured by subjecting a semiconductor wafer to a process such as photolithography, metal deposition or vapor deposition, or etching to form required parts.

[0003] Usually, in order to simplify the manufacturing process, first, a plurality of patterns are repeated using a mask capable of forming a plurality of patterns on the same wafer (here, each of these patterns corresponds to one chip). Is defined, and a series of processes such as etching are performed. Thereafter, the wafer is cut into a plurality of chips.

FIG. 1 shows a typical method of forming a pattern on a wafer. A mask (not shown) used to form a pattern on the wafer 1 is not a mask that forms a single pattern, but a mask that can form a pattern group 3. Pattern group 3
Consists of a plurality of patterns arranged in a matrix (currently, a 2 × 3 matrix as shown in FIG. 1 is often used). During manufacture, such a mask is sequentially exposed to a series of predetermined positions on a wafer while being transported by a precision transport mechanism. Therefore, a pattern group arranged in a 2 × 3 matrix at one predetermined position for each exposure is obtained, and a large number of patterns are uniformly formed on the entire wafer.

The method of forming chips from a wafer is generally as described above.

[0006] During semiconductor manufacture, errors can create defect patterns on the wafer due to errors. Therefore, it is necessary to analyze and evaluate the distribution and the like of the defect pattern on the wafer so as to contribute to the improvement of the yield rate. Such analysis and evaluation are performed in units of one pattern group.

The results of the above analysis have important implications for improving the manufacturing process. For example, on a wafer, when most defect patterns are located at the upper right end in a 2 × 3 pattern group, the defect causes are (1) a problem of the mask itself, and (2) an alignment error of the worktable (work error). The platform may be sloped to the right) or other regular defects may be caused. Such an analysis method is based on exposure pattern recognition (SPR, Shot Pat).
This is referred to as “turn recognition”.

However, the above-described SPR analysis method merely evaluates a manufacturing process based on experience or intuition to determine whether or not defects on a wafer are regularly distributed. This is not a statistical analysis method.

[0009]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, an object of the present invention is to systematically and efficiently analyze a defect pattern of a wafer and contribute to improvement of a manufacturing process. An object of the present invention is to provide an analysis method and an apparatus in a process.

[0010]

According to the present invention, there is provided an analysis method in a wafer manufacturing process for achieving the above object, wherein a wafer having a plurality of pattern groups formed of a plurality of patterns formed for each exposure in a photolithography process. On the other hand, in a method for analyzing a defect pattern in the pattern group, that is, a pattern that does not satisfy a requirement of a certain test, (a) a unique distribution pattern indicating a distribution pattern of the defect pattern in the pattern group is provided. (B) defining a plurality of distribution features correlated with at least one of the unique distribution symbols; and (c) recognizing a defect pattern distribution symbol for each of the wafer pattern groups; Causing the pattern group to correlate with at least one of the eigen-distribution symbols; (D) each defect pattern distribution symbol of the wafer is indirectly associated with at least one of the distribution characteristics based on the correlation between the distribution feature and the unique distribution symbol and the correlation between the wafer defect pattern distribution symbol and the unique distribution symbol. Setting a correlation coefficient for each distribution feature corresponding to the corresponding pattern group, and (e) all correlation coefficients corresponding to the corresponding distribution feature for each distribution feature on the wafer. Is accumulated to obtain the amount of influence by the corresponding distribution feature.

In the analysis method according to the present invention, the defect pattern distribution symbol may be matched or substantially matched with one of the unique distribution symbols or a combination of a plurality of unique distribution symbols for each of the defect pattern distribution symbols. It is characterized in that one or a plurality of eigen-distribution patterns are selected and used as eigen-distribution symbols correlated with the corresponding defect pattern distribution symbols.

Further, a reference diagram of distribution characteristics versus influence amounts may be created from the influence amounts corresponding to the distribution characteristics.

Further, for any one of the pattern groups, if the number of distribution features indirectly correlated with the pattern group is n, the correlation coefficient corresponding to each of these distribution features is 1 / n can be set. The pattern group may be a plurality of patterns arranged in an axb matrix (a and b are natural numbers).

On the other hand, the analysis method in the wafer manufacturing process according to the present invention is a method for analyzing a wafer having a plurality of pattern groups formed by a plurality of patterns formed for each exposure in a photolithography process, including a defect pattern in the pattern group, That is, in an apparatus for analyzing a pattern that does not satisfy a certain test requirement, (1) a first database that records a plurality of eigen-distribution symbols indicating the distribution pattern of the defect patterns in the pattern group; A) a second database for defining a plurality of distribution features correlated with at least one of the unique distribution symbols, and (3) a third database for providing a plurality of recording areas corresponding to each of the plurality of distribution features. And (4) recording a defect pattern distribution symbol relating to the pattern group of the wafer. A fourth database, (5) first linking means for linking each distribution feature of the second database with one or more unique distribution symbols of the first database, respectively, and (6) each link of the fourth database. Second linking means for linking each of the defect pattern distribution symbols with one or more unique distribution symbols of the first database;
(7) Each defect pattern distribution symbol is indirectly linked to the corresponding defect pattern distribution symbol for each defect pattern distribution symbol based on the linking by the second and first linking means between the fourth database and the second database. Analysis means for determining a distribution feature and setting a correlation coefficient for each distribution feature indirectly linked to the corresponding defect pattern distribution symbol; and (8) all the distribution features corresponding to the corresponding distribution feature for each of the distribution features. Calculation means for obtaining the amount of influence by accumulating the correlation coefficients of the above and storing the amount of influence in the corresponding recording area of the third database.

[0015] In the analyzing apparatus of the present invention, the second linking means also has an image analyzing means, and the image analyzing means includes a unique distribution symbol in the first database and a defect in the fourth database. Analyze the pattern distribution pattern,
One or a plurality of eigen-distribution patterns are selected from the first database so that each of the defect pattern distribution symbols matches or roughly matches a combination of one eigen-distribution symbol or a plurality of eigen-distribution symbols. You may comprise so that it may be set as the fixed distribution design linked | related with a defect pattern distribution design.

[0016] Further, there may be provided output means for outputting an influence amount corresponding to the distribution feature from the third database. Further, the output means may be any one of a display, a printer, and a television. Further, the analyzer of the present invention can be formed in a computer.

Further, for any one of the defect pattern distribution symbols in the fourth database, the number of distribution features indirectly linked to the corresponding defect pattern distribution symbol is n.
In this case, each of the correlation coefficients set by the analysis means for each of the distribution characteristics may be 1 / n. In addition,
The pattern group is an a × b matrix (a and b
May be a natural number).

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The structure and operation of an embodiment of the present invention which achieves the above-mentioned objects and solves the problems of the related art will be described in detail with reference to the accompanying drawings.

First Embodiment FIGS. 2A and 2B are views for explaining a first embodiment of the present invention.

Referring to FIG. 2A, in this embodiment, 2 × 3
The analysis method of the present invention is applied to the wafer 100 on which the matrix pattern groups 10 are formed.
The method of the present invention comprises the following steps (for convenience of explanation, the order of execution of each step is as follows, but is not necessarily limited to this order).

First, as shown in FIG. 2B, a plurality of eigen-distribution symbols 20 are defined. Each of these unique distribution patterns 20 indicates how the defect pattern 11 is distributed in one pattern group 10 of the wafer, that is, the distribution form (formable form) of the defect pattern in the pattern group. Is what you do.

Next, a plurality of distribution features are defined. In the present embodiment, as shown in FIG. 2B, how the defect patterns 11 are distributed in the pattern group 10 is classified and displayed based on the distribution characteristics. For example, each distribution feature is defined as all good, upper, upper right, right, lower right, lower, lower left, left, upper left, center, all bad, area bisection, area trisection, etc.
D, UP, RU, RR, RD, DN, LD, LL, L
U, MD, WP, BI, TR, etc., and indicate the distribution form of various defect patterns 11. Then, the eigen-distribution symbols 20 are classified so that these distribution characteristics are correlated with the eigen-distribution symbols 20 on a one-to-one basis or one-to-many basis (indicated by arrows in FIG. 2B). For example,
As shown in FIG. 2B, the distribution feature UP corresponds to the unique distribution patterns 20 (five types) in which the defective symbol is above the pattern group.

Next, the defect pattern distribution symbol is recognized for each pattern group 10 of the wafer 100, and the corresponding pattern group is correlated with at least one of the unique distribution symbols. That is, the wafer 100
, One of the pattern groups 10 is selected, and the defect pattern distribution symbols of the pattern group 10 are compared with the unique distribution symbols 20 one by one. If both are the same, the pattern group 10 is compared with the corresponding unique distribution symbol 20. Make them correlated (for example, indicated by arrows in the figure). When the defect pattern distribution symbol of the pattern group 10 is not in the column of the unique distribution symbol, the unique distribution symbol is set so that the defect pattern distribution symbol of the pattern group 10 substantially (preferably completely) matches a combination of a plurality of unique distribution symbols. (In this embodiment, usually three or less) are selected, and this is set as a fixed distribution pattern correlated with the corresponding defect pattern distribution pattern.

Next, as shown in FIG. 2B, based on the correlation between the distribution feature and the unique distribution pattern 20 and the correlation between the defect distribution pattern and the unique distribution pattern 20 of the pattern group 10, each pattern group 10 ( That is, each defect pattern distribution symbol) is indirectly correlated with the distribution feature. Here, one pattern group 10 can indirectly correlate with one or a plurality of distribution features (indirect correlation).

Next, each of these indirect correlations is regarded as one correlation line, and a correlation coefficient is set for each of the correlation lines (that is, a correlation coefficient is set for each distribution feature corresponding to the corresponding pattern group 10). Set the number of relations.
1, R2, R3, etc. indicate these correlation coefficients). One correlation coefficient indicates the degree of correlation of indirect correlation set for one distribution feature corresponding to one pattern group 10. For example, in this embodiment, assuming that one pattern group 10 corresponds to an n (natural number) correlation line (or a certain distribution feature that is indirectly correlated with the corresponding pattern group), the correlation coefficient corresponding to each of the n correlation lines is Is also 1 / n.

Finally, distribution characteristics (UP, RU, RR, etc.)
For each time, accumulate all correlation coefficients corresponding to the distribution feature,
The resulting numerical value is used as the influence amount due to the distribution feature. This influence amount is determined for all the pattern groups 10 of the wafer 100.
The correlation degree between the defect pattern distribution symbol and the corresponding distribution feature is displayed.

From the influence amounts corresponding to the distribution features as described above, a reference diagram of the distribution features versus the influence amounts as shown in FIG. 3 can be created. In the figure, by comparing the influence amounts between the respective distribution features, it is possible to know in which area the defect pattern 11 is mainly distributed in each pattern group 10. Therefore, it is possible to take measures for improving the manufacturing process.

It is to be noted that a certain eigen-distribution symbol may be composed of a combination of some other distribution symbols. For example, FIG.
In, the symbols corresponding to the distribution characteristics of BI and TR can be composed of combinations of symbols corresponding to UP to MD, respectively. Thereby, the types of symbols can be increased.

Also, for the same wafer, there are two types of curves in the reference diagram of FIG. One is a curve depending on GD to WP and BI and TR distribution characteristics, and one is GD to W
The curve depends only on the P distribution feature. Whether to use BI and TR depends on the needs of manufacturing process analysis.

The present invention is suitable for a wafer composed of an a × b matrix pattern group. Here, a and b are natural numbers, respectively, and are not limited to a 2 × 3 matrix. Second Embodiment An analyzer according to a second embodiment of the present invention is shown in FIG.

According to FIG. 4, the analyzer of the present invention is formed in, for example, a computer 200, and (1) a first database 210 for recording a plurality of eigen-distribution symbols related to the pattern group 10 of the first embodiment; (2) a second database 220 for defining a plurality of distribution features of the first embodiment; and (3) an influence amount described in the first embodiment corresponding to each of the plurality of distribution features. Third database 23 having a plurality of recording areas 231 for recording
0, and (4) each pattern group 10 of the wafer 100
A fourth database for recording the distribution pattern of the defect pattern 11 with respect to each of the distribution patterns of the second database 220 and the twenty-first database 210 by using the method described in (5) the first embodiment. First linking means 2 for linking with one or more unique distribution symbols to be recorded
50 and (6) the defect pattern distribution symbol relating to each pattern group 10 of the fourth database 240 by the method described in the first embodiment.
(Corresponding the defect pattern distribution pattern relating to the pattern group 10 substantially or completely with the combination of the specific distribution patterns), the second correlation means 260, (7) fourth Is indirectly linked to the corresponding defect pattern distribution symbol for each defect pattern distribution symbol based on the linking between the database 240 and the second database 220 by the second and first linking means 260 and 250. The distribution characteristics are determined, and as in the first embodiment,
Analysis means 27 for defining correlation coefficients R1, R2,... For each distribution feature (at least one) indirectly associated with the defect pattern distribution symbol of each pattern group 10.
It consists of 0.

In the present embodiment, for example, one pattern group 10 is indirectly associated with n correlation features, and the correlation coefficient set for each of the n correlation features is 1 / n.

In this embodiment, the analyzer also has (8) calculation recording means 280. The calculation recording means 280 obtains the influence amount by accumulating all the correlation coefficients corresponding to the distribution characteristic for each of the distribution features, and calculates the influence amount in the corresponding database of the third database 230. This is for storing in the recording area 231.

In this embodiment, the second linking means 260 also has an image analysis means 261 (for example, an image analysis program). By executing the program, the defect pattern distribution pattern of the pattern group 10 of the fourth database 210 substantially (preferably, completely) matches the combination of the plurality of characteristic distribution patterns. Column), one or a plurality of eigen-distribution symbols are selected, and are used as eigen-distribution symbols associated with the corresponding defect pattern distribution symbol.

Further, in this embodiment, the analyzer is the third device.
And an output unit 290 for outputting an influence amount corresponding to the distribution feature from the database 230. The output means 230 is, for example, a computer display, a printer, or a television.

Although the present invention has been presented as in the above embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications within the spirit and scope of the present invention.

[0037]

As described above, the use of the method and apparatus of the present invention makes it possible to systematically and efficiently analyze the defect pattern of a wafer so as to contribute to the improvement of the manufacturing process.

[Brief description of the drawings]

FIG. 1 is a diagram showing a typical wafer and a pattern group formed on the wafer.

FIG. 2a is a view showing a wafer and a pattern group formed on the wafer according to the analysis method of the present invention.

FIG. 2b is a diagram showing definitions of various symbols used in the analysis method of the present invention.

FIG. 3 is a reference diagram of a distribution feature versus an influence amount by the analysis method of the present invention.

FIG. 4 is a block diagram showing the structure of the analyzer of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1, 100 Wafer 3, 10 Pattern group 11 Defect pattern 20 Unique distribution pattern 200 Computer 210 First database 220 Second database 230 Third database 240 Fourth database 250 First linking means 260 Second Coupling unit 261 Image analysis unit 270 Analysis unit 280 Calculation recording unit 290 Output unit R1, R2, R3 Correlation coefficient code

Claims (12)

[Claims] 1. A wafer having a plurality of pattern groups each consisting of a plurality of patterns formed for each exposure in a photolithography process does not satisfy a defect pattern in the pattern group, that is, a certain test requirement. A method for analyzing a pattern, wherein: (a) defining a plurality of eigen-distribution symbols indicating a distribution form of the defect pattern in the pattern group; and (b) a distribution feature correlated with at least one of the eigen-distribution symbols. (C) recognizing a defect pattern distribution symbol for each pattern group of the wafer so that the corresponding pattern group is correlated with at least one of the unique distribution symbols; and (d) ) Correlation between the distribution feature and the unique distribution pattern, and the defect pattern distribution of the wafer Based on the correlation between the pattern and the unique distribution pattern, each defect pattern distribution pattern on the wafer is indirectly correlated with at least one of the distribution features, and each of the distribution features corresponding to the corresponding pattern group. Setting a correlation coefficient; and (e) accumulating all correlation coefficients corresponding to the distribution characteristic for each distribution characteristic in the wafer to obtain an influence amount by the distribution characteristic. Analysis method. 2. For each of the defect pattern distribution symbols, one of the columns of the characteristic distribution symbol is matched so that the defect pattern distribution symbol matches or substantially coincides with one of the unique distribution symbols or a combination of a plurality of unique distribution symbols. 2. The analysis method in the wafer manufacturing process according to claim 1, wherein a plurality of selected patterns are selected as a unique distribution pattern correlated with the corresponding defect pattern distribution pattern. 3. The analysis method in a wafer manufacturing process according to claim 1, wherein a reference diagram of distribution characteristics versus influence amounts is created from the influence amounts corresponding to the distribution characteristics. 4. For any one of the pattern groups, if the number of distribution features indirectly correlated with the pattern group is n, the correlation coefficient corresponding to each of these distribution features is 1 / 2. The method according to claim 1, wherein n is set to n. 5. The method according to claim 1, wherein the pattern group is a plurality of patterns arranged in an a × b matrix, and a and b are natural numbers, respectively. 6. For a wafer having a plurality of pattern groups each consisting of a plurality of patterns formed for each exposure in a photolithography process, a defect pattern in the pattern group, that is, a certain test requirement is not satisfied. An apparatus for analyzing a pattern, comprising: (1) a first database for recording a plurality of eigen-distribution symbols indicating a distribution pattern of defect patterns in the pattern group; and (2) a correlation with at least one of the eigen-distribution symbols. (3) a third database for providing a plurality of recording areas corresponding to each of the plurality of distribution features, and (4) a pattern group of the wafer. A fourth database for recording a defect pattern distribution symbol, and (5) a second database. (1) first linking means for linking each of the distribution features with one or a plurality of unique distribution symbols of the first database; and (6) linking each of the defect pattern distribution symbols of the fourth database with one of the first databases. Or a second linking means for linking a plurality of unique distribution symbols, (7) a defect pattern distribution pattern for each of the fourth and second databases based on the linking by the second and first linking means. Analyzing means for determining a distribution feature indirectly associated with the corresponding defect pattern distribution symbol and setting a correlation coefficient for each distribution feature indirectly associated with the corresponding defect pattern distribution symbol, 8) The amount of influence is obtained by accumulating all the correlation coefficients corresponding to the distribution feature for each of the distribution features, and the amount of influence is calculated by the third data. Analyzer of wafer fabrication process comprising the calculation storage means for storing the corresponding recording area of the database. 7. The second linking means further includes an image analysis means, which analyzes a unique distribution symbol in the first database and a defect pattern distribution symbol in a fourth database. One or a plurality of eigen-distribution symbols are selected from the first database so that each of the defect pattern-distribution symbols matches or roughly matches a combination of one eigen-distribution symbol or a plurality of eigen-distribution symbols. The analyzer according to claim 6, wherein the analyzer is a fixed distribution pattern linked to the pattern distribution pattern. 8. The analysis apparatus according to claim 7, further comprising an output unit for outputting an influence amount corresponding to the distribution feature from the third database. 9. The analyzer according to claim 8, wherein the analyzer is formed in a computer. 10. The analyzer according to claim 9, wherein the output unit is one of a display, a printer, and a television. 11. In the case where the number of distribution features indirectly associated with a corresponding defect pattern distribution symbol is n for any one of the defect pattern distribution symbols in the fourth database, the analysis means may determine the distribution pattern of each of the distribution patterns. 7. The analyzer according to claim 6, wherein each of the correlation coefficients set for the features is 1 / n. 12. The analyzer according to claim 6, wherein the pattern group includes a plurality of patterns arranged in an a × b matrix, and a and b are each a natural number.